The use of social networks as a predictive tool to predict the outcome of an election can generate controversy; however if you have a methodology that tries to equate the extracted data as if they were obtained through a conventional survey, that is to say using weighting factors more than what usually should be done, polarity and relevance of each tweet, can make it a very reliable technique in light of the results obtained; the proposed methodology was applied in the presidential election of Ecuador on February 19th, 2017.

Massive Open Online Courses and e-learning represent the future of the teaching-learning processes through the development of Information and Communication Technologies. They are the response to the new education needs of society. However, this future also presents many challenges such as the processing of online forums when a huge number of messages are generated. These forums provide an excellent platform for learning and connecting students of the subject, but the difficulties in following and searching the vast volume of information that they generate may produce the opposite effect.

This paper focuses on the contribution of temporal relations inference and distributional semantic models to the event ordering task. Our system automatically builds ordered timelines of events from different written texts in English by performing first temporal clustering and then semantic clustering. In order to determine temporal compatibility, an inference from the temporal relationships between events –automatically extracted from a Temporal Information Processing system– is applied.

For the healthcare sector, it is critical to exploit the vast amount of textual health-related information. Nevertheless, healthcare providers have difficulties to benefit from such quantity of data during pharmacotherapeutic care. The problem is that such information is stored in different sources and their consultation time is limited.

La investigación en enfoques multidominio innovadores y flexibles puede ser un paso significativo en el área de Generación del Lenguaje Natural. En este sentido, el objetivo de este artículo es presentar un enfoque estadístico centrado en la fase de realización. Este enfoque permite la generación de oraciones que cumplan un propósito dado por una “característica semilla” de entrada, la cual se encargará de guiar el proceso de generación.

En este trabajo se propone el análisis de técnicas adecuadas para el dise˜no y desarrollo de un enfoque de generación de resúmenes multigénero, tomando como partida distintas fuentes de datos pertenecientes a distintos géneros textuales. El objetivo principal es combinar todos estos géneros y producir un resumen abstractivo, es decir un nuevo texto coherente que capte las ideas fundamentales sobre un tema recogidas en las fuentes de datos originales.

El reconocimiento y la clasificación de entidades nombradas (RCEN) es clave para muchas aplicaciones de procesamiento de lenguaje natural. Sin embargo, la adaptación de un sistema RCEN resulta costosa, ya que la mayoría solo funcionan adecuadamente en el dominio para el que fueron desarrollados. Considerando esta premisa, se evalúa si un sistema de clasificación de entidades nombradas basado en perfiles y aprendizaje automático obtiene los mismos resultados independientemente del dominio del corpus de entrenamiento.

Con el fin de desarrollar un sistema de prevención del suicidio en la red, se ha compilado y anotado un corpus piloto de mensajes de ideación suicida extraídos de las redes sociales. Los textos se han obtenido tanto de la Web como de la Deep Web, y se han seleccionado textos escritos tanto en español como en inglés. Para caracterizar semánticamente cada mensaje, éstos han sido anotados según su relación con el fenómeno suicida (pro-suicida, instigador, anti-suicidio, etc.).

This demo showcases a Web application that allows users to easily obtain a summary that is automatically generated taking into account the information provided by other users on the Internet. The application integrates several types of summaries, outlining the most relevant positive opinions, negative and both about restaurants and hotels. In addition, it provides multimodal information, such as graphics, maps or pictures. The results obtained from an on-line questionnaire conducted with real users reveals the potential and usefulness of such an application in the current society.

This paper presents a Named Entity Classification system, which uses profiles and machine learning based on [6]. Aiming at confirming its domain independence, it is tested on two domains: general - CONLL2002 corpus, and medical - DrugSemantics gold standard. Given our overall results (CONLL2002, F1 = 67.06; DrugSemantics, F1 = 71.49), our methodology has proven to be domain independent.